Is Nigeria’s Electoral Voters Data ‘Fake’, As Guardian UK Alleges?, By Simbo Olorunfemi
A deliberate, systematic disinformation exercise is going on to misinform, deceive and mislead the gullible. As to be expected, some are lapping this up, sharing, defending and deploying the offensive information in furtherance of the narratives that suit them. But even for the perceptive, there is the need for extra vigilance…
In the early hours of last Monday morning, just as I was about going to bed, I chanced upon a report by The Guardian UK, shared by two of my friends. It was titled “Nigerian voter data ‘statistically impossible’, analysis shows”. Guardian has now changed the title of the report, just as it has been fiddling with the main story, editing and re-editing it.
With just a casual read-through, I could immediately tell that something was not right with the report. I sensed mischief. So, to my friends, I said: “I have serious reservation about this story. How do you claim that something is statistically impossible, yet with no single reference to data to buttress the claim? What is the point? That the figures were artificially boosted in the same ratio across all the states? How daft an assumption to make. So, with the way the technology deployed is supposed to work, ghosts that have been inserted into the register will vote spiritually to influence the outcome? I guess this is one of those reports deliberately planted in the orchestration to discredit the results. It has too many gaps. Why are they updating the story on the strength of someone contesting their claim? I hope I can squeeze out some time to process the data to see what the scatter diagram will look like.”
I could no longer go to bed, I had to put some time into digging further, interrogating the claims and engaging the voter data for the purpose of establishing correlation and generating my own scatter graph to be placed side-by-side theirs. I eventually went to bed at 3 a.m. But by 8 a.m, I was back at my desk. The first shock for me was that while I was asleep, The Guardian UK had gone to work on the story, removing the statistics it had put up as basis for making that wild claim about Nigeria’s voter data.
But the original story could still be tracked to other sources which had run with it, including our own The PUNCH newspaper, which had slapped it on its pages, even with such a brazen claim that the voters register of the Independent National Electoral Commission (INEC) had been cooked up, if we would call the allegation by the The Guardian UK by its right name. The Guardian UK makes two major claims:
First: Analysis of the INEC data of new voters registered between 2017 and 2018 reveals that “for each of the country’s 36 states and its capital… INEC has increased the number of new registered voters by almost exactly the same percentage across all states.”
“Plotted on a scatter line graph, there was a 0.99 correlation across all the states, without a single outlier. According to three separate data analysts, the parity cannot be a coincidence, and two of them called it a “statistical impossibility”. “Only God works that closely,” one said. The correlation is a “statistical impossibility” and does not reflect Nigeria’s demographic changes, according to data analysts working with the Guardian.”
… is it not often said in statistical analysis that correlation is not causation and that to assume so could lead to correlation fallacy? If indeed correlation had been established, would it not have been lazy and unprofessional to immediately assume causation, without a consideration of many other factors that could have precipitated correlation?
“On average, voter registration in each state increased by 2.2 per cent between April 2017 and January 2018, and by 7.7 per cent for the whole registration period ahead of Saturday’s election.”
Second: The paper also looks back to 2015 and reveals thus: “Additional data seen by the Guardian also shows irregularities in registration for the 2015 election, until now considered to have been free and fair. An analysis of separate figures shows that manipulation may have happened in favour of Buhari’s party, which was running in opposition to Goodluck Jonathan’s People’s Democratic Party. A clue may have been dropped last July when the INEC, perhaps inadvertently, publicly referenced a different set of results from the one on which Buhari’s victory was based. Both documents showed 29.4 million votes were cast. But according to the original results, 31.7 million accredited voters participated in the election, whereas in the second set of results, that figure dropped to 23.6 million.”
These are rather serious claims and allegations. There is no way one could come across this and simply roll onto the other side of the bed. One would have to dig. This I did. What did I find out?
Starting with the first claim, my first instinct was to establish correlation, dumping some of the data from 2015 and 2018 into SPSS to process, analyse and generate a scatter graph from. I immediately found out that the level of correlation for the data inputted was at a significant level. But then, I pulled back to ask: Why would it be strange that there is correlation between 2015 and 2018? What significant demographic changes could have occurred?
I went back to the report by the paper, noticing its claim that the said article, “was amended on 15 February to reflect doubts over the way in which analysts who spoke to the Guardian had interpreted new voter registration data for the 2019 election.” The report also incorporates the response of Nonso Obikili, who the paper cites as an economist who told them that the correlation “in itself, it’s not an indicator of any kind of electoral malpractice or manipulation” …and that the correlation in the registration figures across states was “what you would expect, because demographic changes don’t happen very quickly.”
In any case, is it not often said in statistical analysis that correlation is not causation and that to assume so could lead to correlation fallacy? If indeed correlation had been established, would it not have been lazy and unprofessional to immediately assume causation, without a consideration of many other factors that could have precipitated correlation? Except there is an agenda being pursued or a predetermined objective for the exercise, possibly prompting the choice of a scatter graph, among all tools, it is difficult to see the sense in the line pursued by The Guardian UK and its data analysts.
One would have thought that by now Guardian UK would have pulled down this story and promptly apologised to Nigeria and Nigerians for this assault and sought to prove to us that this story was not instigated by forces with a not-too discreet agenda to discredit the 2019 elections, out of fear that its wish might not be in tandem with that of the majority of Nigerians.
But then, not to forget, the paper had, in fact claimed a uniformity in the percentage increase of voter registration across all the states, upon which its claim of statistical impossibility rests. What does the data say? I had only analysed the voter registration figures for a few states to come to the realisation that such a claim can only be resting on shaky, possibly fictitious legs, as there was nothing to suggest that such claims could be valid. I was not surprised that on return to the story, The Guardian UK had edited out those figures.
On the second claim by the paper alleging irregularities in the registration for the 2015 elections, one would have wished that the paper would refer to the data and “separate figure” that it has come by, upon which it comes to this conclusion that “manipulation may have happened in favour of Buhari’s party, which was running in opposition to Goodluck Jonathan’s People’s Democratic party.” This is not to defend Buhari or INEC (unnecessary in my view). But the fact is, apart from the problem with the data, where The Guardian UK leans, in this story, is easy to see from its summation on the incumbent’s record and silence on the opposition’s record or reputation.
What should be of concern is the “additional data” The Guardian UK has come by that it will neither publish or even make a reference to elements from it to buttress its claim. The election results for 2015 have always been in the public domain. They have been subjected to review by the public, the civil society and even the courts. At no point has there been such a claim, as the paper makes, of two sets of results, one reading 31.7 million accredited voters and the other 23.6 million accredited voters, whereas 29.4 million votes were cast. Perhaps The Guardian UK does not know that these data were organically generated from bottom up with different layers of collation and verification. Where did the paper come by these data?
The Guardian alleges that “a clue may have been dropped last July when the INEC, perhaps inadvertently, publicly referenced a different set of results from the one on which Buhari’s victory was based.” Yet there is no evidence presented to back that allegation. The link inserted by the paper as proof takes one to a completely unrelated story about the registration of more political parties. Where and when did INEC publicly reference a different set of results, as Guardian UK alleges?
This is really disturbing. It is not enough that the paper has edited the story, changed the headline to something less offensive, elements of the spurious claims, around which the original story was written, are still there. This misleading and disparaging report has no leg to stand on. It is a shame if the paper pretends not to know that. One would have thought that by now Guardian UK would have pulled down this story and promptly apologised to Nigeria and Nigerians for this assault and sought to prove to us that this story was not instigated by forces with a not-too discreet agenda to discredit the 2019 elections, out of fear that its wish might not be in tandem with that of the majority of Nigerians. It will be unfortunate if Guardian UK will trade its name and reputation in aid of the enemies of Nigeria desperate to have their way, whatever the cost to the Nigerian people.
A deliberate, systematic disinformation exercise is going on to misinform, deceive and mislead the gullible. As to be expected, some are lapping this up, sharing, defending and deploying the offensive information in furtherance of the narratives that suit them. But even for the perceptive, there is the need for extra vigilance, given the subtle nature of the disinformation machinery. The space here is already so toxic, and to deliberately foul the air for a repugnant objective, is a recipe for disaster. But those who are planning for Nigeria to fail can only fail in their quest and machinations.